Systematic data relating to marital, sexual, and social experience were obtained by means of structured interviews with male members of the Baltimore Longitudinal Study on Aging. During the year stepwise multiple regression procedures were used to identify those factors which, alone or in combination, best predicted current frequency of sexual expression. The best three-variable model which emerged included: time comfortable without sex, age at report, and customary frequency of coitus in the first year of marriage. Why these particular proved to have a predictive function and how they may be useful in future inquiry are questions of some importance.